Meta suspends Muse link-sharing after security firm uncovers exposed AI chats
Meta has halted the feature that lets users share Muse AI conversations via links after researchers discovered thousands of those links publicly indexed.
A security analysis by Claroc revealed that over 6,800 Muse conversation links were exposed through public DNS lookups, making private AI chats accessible to anyone who discovered the URLs. The link format combines the conversation's title with a random identifier, and the hostname often hinted at the page's content, facilitating easy reconstruction of the links. Examples of exposed data included a marketing firm’s revenue figures, a small business owner's loan request, a company's full-year staff calendar, and interview notes for an AI lab candidate.
Meta confirmed that it has temporarily disabled the Muse sharing capability while it evaluates the report and explores changes to prevent future exposure. Tracy Clayton, a spokesperson for Muse, said the pause is intended to address the security concerns raised by the report.
Why it matters
Exposed AI chat links could leak confidential business and personal information, highlighting privacy risks in emerging tech features.
In this story
